o Predicted risk scores of 10 chronic diseases in United States with NHANES data, extracted demography, symptom, lab/examination indicator, comorbidity features, built a cost-sensitive model pipeline (MetaCost, LR, XGBtree), visualized top 5 important features individually by Python
Final Risk Score Prediction Model Presentation
Model Pipeline
An example of predicting COPD (Chronic obstructive pulmonary disease)
o Lifebook Platform Architecture Proposal: Designed proposal with three modules: data acquisition, data integration and storage, data analysis and service, and planned the function setting, community building, privacy and profit model.
Data mining projects include analysis of patient, provider and insurance claim data.